Integrating uncertainty into a language for knowledge based systems

Abstract

Fm is an object oriented language designed to serve as a testbed for experiments in the development of conceptual structure in continuous domains. Possibilistic truth representation is fully integrated into a language for building knowledge based systems offerring support for object, rule, and data-access based programming styles as well as the more traditional procedural form. A prototype implementation of Fm has been written in Franzlisp and has been used to construct two simple expert systems, one a data structure consultant and the other a knowledge based automobile driver. This paper describes the facilities provided in Fm for representing and reasoning with uncertain information.